When discount factors within the household differ, implementing the ex-ante Pareto efficient consumption allocation requires the ability to adhere to binding intertemporal contracts. In the absence of commitment, the availability of a “private” savings technology (a device that is only accessible by a single owner) may incite individuals to take costly strategic savings action in order to manipulate the time path of consumption. This paper presents a model that formalizes this idea and derives several testable theoretical implications. In particular, households where husbands and wives are well matched in terms of time preference should make greater use of joint (public) accounts, less use of individual (private) accounts, and make more efficient investment choices as compared to their poorly matched peers. The model informed the design of a field experiment where married couples in rural Kenya were given the opportunity to open joint and individual bank accounts at randomly assigned interest rates. The behavior of individuals in the experiment is inconsistent with ex-ante Pareto efficiency and a variety of alternative models of intrahousehold resource allocation, but consistent with the proposed model of strategic savings. Savings misallocation due to strategic behavior may be substantial: in the experiment poorly matched couples forgo at least 64 percent more interest than well matched couples.